Papers 80 001-312 13/12/06 10:53 Página 149

Papers 80, 2006 149-169

Homo Sociologicus: Neither a Rational nor

an Irrational Idiot
Raymond Boudon
Université de Paris-Sorbonne (Paris IV).
Centre d’Études Sociologiques de la Sorbonne
rboudon@noos.fr

Abstract

As the social sciences deal with macrophenomena which are caused by individual human
actions, they have to use a theory of human behaviour. They use basically three types of
theory: the rational-utilitarian theory, the causalist theory which sees behaviour as caused
by social, cultural or biological forces, and the «rational psychology» in Nisbet’s sense. The
three theories are important in the sense that they have shown their capacity to explain
convincingly puzzling phenomena. Type I and II theories have been claimed to be poten-
tially general. They are not since there are familiar phenomena they are unable to explain.
Type III can by contrast be claimed to be general. Amartya Sen has made the point that
the rational-utilitarian theory treats human beings as rational idiots. Type II treats them
as irrational idiots. These aggressive metaphors draw the attention on the fact that human
beings answer the situations they are confronted to by devising systems of arguments which
they perceive as strong: by being cognitively rational.
Key words: rational choice, cognitive rationality, axiological rationality, theories of behav-
iour.

Resumen. El homo sociologicus: ni idiota racional ni irracional

Puesto que las ciencias sociales estudian macrofenómenos causados por acciones humanas
individuales, tienen que utilizar una teoría del comportamiento humano. Utilizan básica-
mente tres tipos de teorías: la teoría racional-utilitaria, la teoría causal que ve el comporta-
miento como causado por fuerzas sociales, culturales o biológicas, y la «psicología racional»
en el sentido de Nisbet. Las tres teorías son importantes en el sentido de que han demostra-
do su capacidad para explicar convincentemente fenómenos enigmáticos. Se ha pretendido
que las teorías del tipo I y II eran potencialmente de aplicación general. No lo son, dado que
hay fenómenos familiares que son incapaces de explicar. Las del tipo III, por el contrario,
pueden pretender generalidad. Amartya Sen ha argumentado que la teoría racional-utilita-
ria trata a los seres humanos como idiotas racionales. Las teorías del tipo II los tratan como
idiotas irracionales. Estas agresivas metáforas llaman la atención sobre el hecho de que los
seres humanos responden a las situaciones con las que se enfrentan ideando sistemas de argu-
mentos que perciben como poderosos: esto es, siendo cognitivamente racionales.
Palabras clave: elección racional, racionalidad cognitiva, racionalidad axiológica, teorías
del comportamiento.
Papers 80 001-312 13/12/06 10:53 Página 150

150 Papers 80, 2006 Raymond Boudon

Contents
A typical analysis: Durkheim analyses Type III model
a correlation Type III model and Weber’s
Three theories of behaviour methodological writings
Type I or II? The importance of the discussion
Is Type I general? References
Choosing Type II instead of Type I?

A typical analysis: Durkheim analyses a correlation

Durkheim (1967[1897]) observes that suicide rates go down when a serious
political crisis appears. They go down in France at the time of the main crises
of the Third Republic: the Dreyfus and Boulanger Affaires. They decline
both in Austria and Prussia during the tension and the war between the two
nations which was to be concluded by the Prussian victory at Sadowa. Why
this counterintuitive correlation? Because, Durkheim suggests, when a seri-
ous national or international crisis appears, people cannot help worrying at
what happens in the outside world and are provisionally deterred from their
personal problems. Once this psychological mechanism is made explicit, our
confidence in the fact that the correlation reflects a genuine causal relation-
ship is increased.
I think this analysis includes an essential methodological teaching: it exem-
plifies a procedure characterizing, not only sociology, but all sciences.
An investigation begins often with some macroscopic observation, as a cor-
relation between two variables, not only in sociology but also, say, in the life sci-
ences. Thus, a study on nutrition is likely to start from the macroscopic obser-
vation that some type of cancer appears as less frequent among people who
eat oil from vegetal rather than animal origin. But, one will be really certain
that the correlation reflects a genuine causal relationship only from the moment
when it will have been possible to describe the elementary mechanisms respon-
sible for the correlation. And, even when the design of experiment is con-
ducted carefully enough, so that there is little doubt that the correlation is due
to a genuine causal relationship, the investigation will still give the feeling of
being unfinished, since the question will remain as to from which elementary
causes the macroscopic causal relationship come from.
So, a general rule of any scientific investigation in any science is that, in
order to explain a macroscopic phenomenon, the analyst should try to identity
the ultimate elementary causes responsible for the observed macroscopic phe-
nomenon. As far as the social sciences are concerned, Weber and Schumpeter
have given this general rule a name: «methodological individualism», since the
elementary causes of any social phenomenon are obviously human individual
actions. Once this evident principle is recognized, crucial questions appear:
how are human actions validly explained? What are their causes? Is it possible
Papers 80 001-312 13/12/06 10:53 Página 151

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 151

to identify these causes? How? These questions raise the more general ques-
tion as to «Which theory of behaviour for the social sciences?».
This question has been and is still intensively debated in the social sciences,
because it is crucial. It seems to me the point can be made that the variety of
answers the social sciences have given and give to this basic question can be
reduced to three basic types.

Three theories of behaviour

Type I. The first Type is illustrated by the so-called «Rational Choice Theory»
(RCT), also called the «expected utility model». According to theorists such
as Gary Becker (1996) or the late James Coleman (1986), RCT should be con-
sidered as the theory of behaviour valid for all social sciences. According to
this model, people do what they do because they think that their action will give
them a maximum satisfaction. The kind of psychology used in the RCT can
be qualified as utilitarian. It can also be called consequential, since it assumes
that people do what they do in consideration of the consequences their action
will generate. This type of theory was developed systematically by Jeremy
Bentham. It has been revivified in the last years.

Type II. According to the second type of theory, behaviour should in most
cases be explained as the effect of psychological, cultural or biological forces:
by socialization effects, by cognitive biases, as the outcome of biological
evolution, as the product of some instinct such as the instinct of imitation,
etc. While these forces can be psychological or cultural, the ultimate causes
underlying them are biological. They cannot in most cases be disentangled by
biology in the present state of the art though. For this reason, the forces thanks
to which they explain behaviour have the status of mere conjectures, except
when they deal with psychological mechanisms known since ever.
The second Type of model is used by many sociologists and many anthro-
pologists, as Clifford Geertz (1984). They do not deny the fact that individuals
have intentions, but they maintain that their intentions are determined by
psychological, cultural or biological forces, i.e. by causes beyond the control
of the individuals themselves. They contend that intentional as well as non-
intentional behaviour should generally be explained by causes beyond the con-
trol of individuals. The psychology used by this model can be called causal-
ist, though this appellation raises the objection that not only forces but reasons
can be causes.

Type III. The third model is used explicitly by Max Weber. But it is also used
implicitly by Tocqueville and even by Durkheim and by many of their follow-
ers. They consider that the cause of an individual action is its meaning in the
mind of the individual. They regard individual actions as being in principle
understandable, as inspired either by understandable motivations or by reasons.
Following Nisbet (1966), I propose to call rational the kind of psychology used
Papers 80 001-312 13/12/06 10:53 Página 152

152 Papers 80, 2006 Raymond Boudon

in this model. This psychology is the psychology developed by Aristotle, and

by the French moralists of the 17th century. In a word, it is no other than the psy-
chology used in everyday life, or in institutional life, in police stations or in
judiciary institutions for instance. This type of rational psychology is used in
the example from Durkheim I mentioned earlier: when the fire burns outside,
I forget for a while grumbling about my personal problems: an understandable
reaction. Tocqueville shows that in 1789 the French people had reasons to be hos-
tile to the clerics which the British people had not. Hence the macroscopic
effect that the French enlightenment was much more anticlerical than the British
one or than the German Aufklärung. Max Weber has explained that the Roman
soldiers had reasons to prefer the monotheistic to the polytheistic cults. This
explains the quick diffusion of Christianity in the first centuries in the Roman
Empire. Neither Tocqueville, nor Durkheim, nor Weber had a use for causal-
ist psychology. They consider that an action, belief, or attitude can be held as irra-
tional, as produced by hidden «forces», exclusively from the moment when all
rational possible explanations have been exhausted.

Of course, the three types I have distinguished are «ideal types»: many
works are eclectic in the sense that they use simultaneously the three types of
psychology. But still, it can be recognized that, say, Clifford Geertz’ analyses
make little use of rational or of utilitarian psychology, while Tocqueville’s,
Durkheim’s or Max Weber’s analyses, say, have no use for the causalist type of
psychology and use essentially rational psychology in the above sense.
This discussion is far from being merely speculative. On the contrary, the
answer given to the question as to which theory of behaviour the social sci-
ences should preferably use determines to a large extent the effectiveness of
their analyses, as I will try to suggest.

Type I or II?
I will illustrate this point by showing that shifting from one of the three mod-
els to another can lead to substitute an illuminating explanation for a ques-
tionable one.
In a study on Vietnam in the twenties of the last century, Popkin (1979)
wondered why all traditional Vietnamese village societies had adopted the una-
nimity rule as the basic constitutional rule. He looked at the literature on the
question as to why this rule is commonly accepted in village societies, in South
East Asia, but also in Africa and elsewhere. In fact the unanimity rule is cur-
rently used in traditional village societies around the world. Popkin noted that
most available explanations of this phenomenon use a model belonging to the
second type of behavioural theory: they assume that people were motivated
to devise and apply the unanimity rule under the effect of cultural forces.
They mobilize generally the assumption that in traditional societies peo-
ple have a weak feeling of their individuality; that they have the impression to
exist only as one part of the group; that consequently the group itself is in
Papers 80 001-312 13/12/06 10:53 Página 153

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 153

their mind the only subject able to legitimate a collective decision; and final-
ly that to them unanimity is, so to say, a criterion ensuring them that the deci-
sion comes from the will of the group as such.
In other words, the current explanation of the phenomenon illustrates the
second type of psychology: people are supposed to favour the unanimity rule
under the effect of impersonal cultural forces. They perceive themselves as a
mere part of the group under the influence of these forces and consequently find
natural to consider that a decision is worth being enforced only when the
group has given an unambiguous verdict.
Two objections, though, can be raised against this type of explanation. First:
that the unanimity rule is another name for the veto right. In societies where
the unanimity rule prevails, every single member has the capacity of saying no.
In that sense, his opinion has more weight than under any other rule, as the
majority rule. Second objection: in traditional village societies operating under
the unanimity rule, a collective decision is often taken in an atmosphere of con-
flict and after a long discussion, as the very notion of palaver indicates. On the
whole, the causalist explanation introduces an assumption which is both heavy,
since it postulates the existence of cultural forces able to implant representations
in the mind of people, and controversial, since it accepts wrongly that individu-
als have less power under the unanimity than under the majority rule. Finally,
it ignores and fails to account for the well-known phenomenon of the palaver
which always appears as soon as the unanimity rule is adopted in any context.
Popkin made the point that Rational Choice Theory provides a much more
easily acceptable explanation. He started from Buchanan’s and Tullock’s (1962)
theory, a theory which explains why in our societies the majority rule is very gen-
erally treated as a basic decision making rule. This derives from the fact that the
organization of a collective decision procedure is exposed to two types of costs.
If it is decided that all members of the group should agree before a decision
is considered as lawful, the time used in the decision-making process will like-
ly be long. At the other extreme, if a «dictator» were appointed in the sense
that his decisions would be treated as lawful, the collective decision process
would be quick. In other words, the higher the proportion of people in the
group whose agreement is requested before a decision is considered as lawful,
the longer the time taken to reach a decision. On the other hand, the high-
er the proportion of people in the group whose agreement is requested before
a decision is considered as lawful, the smaller the proportion of people exposed
to the risk of seeing a decision enforced with which they would disagree. The
theory assumes furthermore that the two curves representing the two kinds of
costs as a function of the number of people whose agreement is required, are
convex. This assumption derives from the consideration that, when all agree on
a decision except a last opponent, convincing him will likely be more difficult
than convincing less stubborn opponents. So, it seems plausible to assume that
the curve relating the number n of people whose agreement is required to the
first type of cost, i.e. the length of the decision process, is a monotonous increas-
ing convex function of n. As to the other cost, similar reasons lead to assume
Papers 80 001-312 13/12/06 10:53 Página 154

154 Papers 80, 2006 Raymond Boudon

that the curve representing it as a function of n is convex. Under these assump-

tions, the sum of the two costs will in broad conditions be minimum when n
is equal to half the total number of members. This explains why the majority
rule is the rule normally accepted and considered legitimate in many situa-
tions.
But there are also circumstances where it is not, even in our societies.
Thus, by contrast with the adoption of an ordinary new law, a change in the
constitutional law will require, say, a two third majority because, there,
the time curve is moved to the right: given the importance of the decision,
one wants to be sure that it is not imposed to a too great number of people
even if the decision takes time. As to the decision rule on the battle field, it
is «dictatorial» in all armies, because here the time curve is moved to the left,
since taking too much time to reach a decision may be fatal in a war situa-
tion.
Popkin proposes to consider that the question as to why village societies
use generally the unanimity rule is a special case of this general theory. The
Vietnamese traditional village societies are characterized by an economic regime
of the subsistence type. There is no economic surplus; the production covers
at most the local consumption. If the majority rule, say, was accepted, nothing
would prevent for instance the group from forbidding gleaning. Now, the
poorest slice in the population lives from gleaning. So, under the majority
rule, some members of the society would be heavily threatened. Such a rule
would consequently not be considered legitimate. In other words, the curve
representing the second type of costs, the «external costs», is moved far to the
right in the case of this type of society. These costs are very heavy except if all
agree. On the other hand, time has much less value in traditional village soci-
eties than in modern societies, since the subsistence economy requires from
the producers only a small part of the time available to them, so that the length
of time spent to take a decision is much less important than in modern soci-
eties.
Of course, it cannot be supposed that the members of the considered
societies have effectively this theory in mind. Still, one can legitimately sup-
pose that they have in mind a rough intuitive version of the theory. They know
well that a minority lives from gleaning and that a majority could decide to
outlaw this traditional institution for instance. Leo Moulin (1953) has shown
that the majority rule was invented in the convents of the Middle Ages on the
basis of considerations which anticipated literally on Buchanan and Tullock’s
theory. On the whole, Popkin’s theory is much more easily acceptable than its
competitor: it includes no heavy assumption, no contradiction with data, no
arbitrary assumption; it is congruent with the relevant data.
So we have here a case where shifting from a Type II model to a Type I,
from a model interpreting people’s behaviour as the effect of «cultural forces»
to a model interpreting it as rational in the sense of the RCT appears as a def-
inite scientific progress. Given the criteria generally used to evaluate a theory,
the RCT appears as much better than the alternative one.
Papers 80 001-312 13/12/06 10:53 Página 155

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 155

Is Type I general?

Should we consider the behavioural theory of Type I as general, though? My

answer is no. For the reason that a number of phenomena appear as reluctant
to be explained by this model, in spite of the sustained efforts made to explain
them by Type I theories.
The case of voting is illuminating in this respect. RCT assumes that peo-
ple’s actions are always motivated by the objective of generating some out-
come. RCT assumes, in other words, an instrumental or consequential theo-
ry of action: that actors are motivated by the consequences they imagine they
are going to produce.
Now, no single vote is likely to have any effect on the outcome of the poll.
Why do people still vote? A possible answer is that they tend to have a biased
view as to the actual importance of their vote. But in that case it is necessary
to explain the origin of this bias. Does it come from some psychological force
that would lead most people to believe falsely that their vote has an impact on
the outcome while it has actually no impact? I doubt that such a theory has
ever been taken seriously. Though some people may have a wrong view of the
influence of their ballot, it is very unlikely to suppose that most voters have.
An alternative theory would be that people still bet that their ballot can
have an influence (Ferejohn et Fiorina, 1974). The 2000 American presiden-
tial election has shown that it can. But even in this very particular case the
probability of a particular ballot having an influence was a priori close to zero.
So that, if we want to elaborate on this assumption, we have to suppose that the
utility to most people of the election is so high that it compensates, so to say,
the infinitesimally weak probability of any ballot being pivotal. This is the
type of argument Pascal used when he tried to convince the atheist that he
should bet that God exists. Since not believing in the existence of God if God
actually exists would condemn the atheist to damnation, it would be rational
for him to believe in God even if the probability that God exists would be as
close to zero as wished. As a mathematician, Pascal plays in this famous argu-
ment with the mathematical paradox that the infinite multiplied by a quanti-
ty as close as wished to zero without being zero is equal to the infinite. But
are we ready to assume that the eventuality of candidate X being elected as
president rather than Y appears to all voters as important as spending the eter-
nity in the flames of Hell?
Other theories belonging to Type I have been tried, such as the theory
which assumes that abstention has a cost. The theory is highly speculative,
though. In some areas and circumstances, people may want effectively to avoid
the eventual reprobation induced by abstention because they are under the
control of their neighbours. In others, they are not. Thus, in Paris, nobody
knows whether I have voted or not.
Many other theories could still be mentioned, which attempt at recon-
ciling RCT with the fact that a non negligible number of people do actually
vote.
Papers 80 001-312 13/12/06 10:53 Página 156

156 Papers 80, 2006 Raymond Boudon

Since none of the theories belonging to Type I appears as really convincing,

they lead to the conclusion that most people see that voting is a non conse-
quential behaviour and that their vote is not going to make the difference.
Common sense tells us that people vote because they have a positive attitude
toward democracy and hold elections as an essential ingredient of democracy,
so that they vote if they think, say, that X is definitely better than Y. This expla-
nation takes us away from RCT. But it is a good theory in the sense that all
the statements it includes are acceptable and that they lead moreover to con-
sequences congruent with data. In particular, it can explain satisfactorily the vari-
ation in the rate of participation: it predicts for instance that the rate of par-
ticipation will generally be lower in the case where many citizens have the
impression that none of the candidates offered to them is palatable.
An argument often heard on RCT is that it is better because it allows
expressing behaviour by simple equations. As biology, the most active and
influential science to day, however shows, equations are neither a necessary
nor a sufficient sign of a scientific approach. The time when mechanics were
considered as the model of all sciences is over. The aim of science is to describe
reality as it is, not necessarily to put it into equations. Now, no voter will rec-
ognize himself in the theories proposed by RCT on voting.
Voting behaviour is not a singular case. RCT fails in many other exam-
ples, as the ultimatum game. In this game, A is asked to propose as he wishes
to share 100 € between himself and B. B can only accept or reject A’s propos-
al. If B accepts A’s proposal, each receives the part of the 100 € A proposed to
grant respectively to A and B. If B refuses A’s proposal, the 100 € remain in
the experimenter’s pocket. If RCT were valid here, A would propose some-
thing like «70 € for me, 30 € for B», since B would accept and A would get a
satisfactory outcome from his proposal. In fact, A proposes in most circum-
stances equal sharing.
In the same spirit, Frey (1997) relates an interesting experiment where two
groups are proposed to decide whether they would accept nuclear waste on
the territory of their community. The local authorities would be proposed a
monetary compensation in the case of the first group. No compensation would
be proposed in the case of the second group. Against the prediction of RCT, the
rate of people accepting the nuclear waste is greater in the case where no com-
pensation is proposed.
These counter examples are taken among many possible other examples.
They are probably sufficient to show that RCT cannot be held as general for
the obvious reason that many actions are not inspired by their utility to the
actor, while other actions appear even as not consequential. In other words,
RCT is valid exclusively when restrictive conditions are met.
Amartya Sen has once said that the social actor as described by the RCT
is a «rational idiot». He meant that a real human being whose constitution
would correspond to the postulates describing the homo œconomicus, the social
actor of RCT, should for instance not vote. As a matter of fact, I know a col-
league who, as a true believer in RCT, actually does not vote. I always won-
Papers 80 001-312 13/12/06 10:53 Página 157

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 157

dered whether he was serious or joking. The social actors of RCT are «rational
idiots» in the sense that in a prisoner’s dilemma game they would easily gen-
erate a catastrophe. Axelrod (1984) has tried to make them a little less idiot
in the case of repeated prisoner’s dilemma games in the sense that in that case,
they can play the tit-for-tat strategy, i.e., offer the other player to cooperate.
If the other accepts and if none of them is tempted by the short term benefits
that a treason strategy would generate, the catastrophe will be avoided. In real
repeated prisoner’s dilemma games it can occur though that actors cannot use
the axelrodian strategy, as illustrated by the case of the arm race between the US
and USSR in the second part of the 20th century. Here, the two actors could
not help behaving as rational idiots and went on a long time with the arm race
until a brilliant innovation brought the arm race to an end: The Strategic
Defence Initiative, the Star War. It had the effect that the other player could not
follow. By so doing, the first player was not «rational», but rather «intelligent»,
in the sense that he destroyed the repeated prisoner’s dilemma game. And he
was able to do so because he had built a good theory of the situation: the other
player is exhausted, if a program of arm development is devised which would
appear to him as too costly, he would have no other choice than to propose
to end the game.
The difference between a rational and an intelligent actor is that the lat-
ter has a good theory of the situation and of the ways able to modify the sit-
uation. He is «rational» in the sense, not of RCT, but in the sense where a sci-
entist is rational. He is cognitively rational.

Choosing Type II instead of Type I?

The failures of RCT are probably to a large extent responsible for its unpop-
ularity among sociologists. But this unpopularity is also due to the fact that
RCT contradicts their favourite model. For many sociologists and anthropol-
ogists accept the idea that explaining behaviour means identifying the psy-
chological, cultural or social forces causing behaviour.
Thus, Clifford Geertz states in an influential article that sociologists and
anthropologists should follow literally Montaigne’s teaching, when he claims
that, as norms and values are different through cultures and as people tend in
all cultures to find them as valid, people believe in the norms and values dom-
inant in their society under the action of these forces which sociologists and
anthropologists call socialization effects.
Socialization effects do exist beyond doubt. But they are too easily evoked
by sociologists. First of all, their existence is generally demonstrated in a circular
fashion. Then, socialisation often does not work. Thus, Weber (1988 [1920-
21]) explains that in the Roman Empire, most roman civil servants and sol-
diers had been educated in the traditional polytheistic roman religion. When
monotheistic cults appeared in the Roman Empire coming from the Orient, as
the Mithra cult and then Christianity, they were strongly attracted by them. So,
socialization apparently did not work properly in this case. Simmel (1892)
Papers 80 001-312 13/12/06 10:53 Página 158

158 Papers 80, 2006 Raymond Boudon

observed on his side that socialization can easily have opposite effects: a per-
son who has had a hard youth can as well become egoistic as generous; a person
educated liberally as a youngster can educate his own children in a liberal or
authoritarian fashion.
Behaviour is also often explained in the social sciences by psychological
forces. Thus, cognitive psychologists have observed that people consider as
evident solutions to simple mathematical or statistical problems which are false
in the eyes of the mathematician or of the statistician. In many experiments,
people appear as underestimating or overestimating probabilities, as seeing a cor-
relation where there is none, etc. Psychologists explain this phenomenon by
evoking psychological biases. In most cases, the existence of these biases is
grounded in a circular fashion by the effects they are supposed to explain,
though. Sometimes, an effort is made to guess where these biases come from.
Thus Tooby and Cosmides (1992) claim that the biases in statistical infer-
ence they registered in some experiments come from the hunting experiences
of men in the early stages of mankind. Such assumptions are highly conjec-
tural. Moreover, even if they could be empirically tested, they would explain
exclusively some particular biases. They explain the under- or overestimation of
probabilities, but not for instance that people in some experiments see a cor-
relation where there is none. Still, these writers have well perceived that the
biases themselves needed to be explained.
Dawkins (1976) theory of memes is fashionable today. Genes explain why
some biological features are transmitted from one individual to another.
Dawkins propose to introduce the notion of cultural genes, which he calls
memes, and to explain cultural evolution by the fact that memes are transmit-
ted from one individual to another, in the same way as biological evolution is
explained by the fact that genes are transmitted from one individual to anoth-
er. Dawkins theory develops actually in a new vocabulary a variant of the clas-
sical theories which propose to explain behaviour by socialization effects, or
by an imitation instinct, as the theory developed by Tarde in his Laws of imi-
tation. The attractiveness of Dawkins’ theory seems mainly due to the fact that
it took benefit of the success, influence and prestige of the well established
theory of biological evolution and that it rewords theories popular among soci-
ologists and anthropologists.
The type II model advertised by Geertz, by Dawkins and by many sociol-
ogists and anthropologists according to whom behaviour should be explained
by some psychological, biological or cultural forces is valid in many circum-
stances. But, as the Type I model, it is not general.
In some cases it works beyond doubt. To take an obvious example: when
Damasio (1994, 2003) observes that some lesion in the brain provokes a
brutal change in the behaviour of an individual, we have little doubt that
the change in behaviour is due to the lesion in the brain. Thus, Damasio
describes the case of a sociable, open, lively and pragmatic mining engineer
who made a mistake while he manipulated an explosive, had his brain dam-
aged and became whimsical and egocentric. In that case, we have no doubt
Papers 80 001-312 13/12/06 10:53 Página 159

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 159

that the change in the state of the brain is the cause of the change in the
behaviour.
We are obviously very far from this unambiguous situation when a soci-
ologist, an anthropologist or a psychologist explains the behaviour of adults
by some experience in their childhood or by the effect of socialization, say.
Here, the cause is in most cases conjectural. As a matter of fact, the concept
of socialization is in most cases descriptive rather than genuinely explanatory.
When a sociologist states that Roman peasants believed in many gods because
they were socialized in the traditional polytheistic roman religion, this amounts
to saying that, as their parents, they were themselves educated in this religion.
This statement registers a fact rather than a causal relationship. Moreover, as I
reported earlier, this case raises the question as to why socialization worked
properly as far as Roman peasants were concerned, but not with the soldiers and
the civil servants.
Or, to come back to the examples of application of the Type II model in cog-
nitive psychology, saying that a subject underestimates some probability or sees
a correlation where there is none under the effect of a bias, because he is exposed
to some bias, does not say more than saying that he is mistaken. The notion of
bias has the status of a stenographic descriptive notion. It could cease to be a mere
word if it could be shown that biological evolution, say, has produced a wiring
of the brain explaining the bias. Otherwise we are left with a black box.
Generally, words like meme, bias, frame, socialization effects, habitus, social
representation and the other concepts of the same type are descriptive rather
than explanatory. This point was already convincingly made by Popper (1976)
in the case of the concept of frame. Such words postulate conjectural biolog-
ical or evolutionary processes which take the social scientist beyond his realm
of competence.
At this point of our discussion, we can conclude that, as shown by Popkin’s
example and many others, models of Type I can lead to illuminating theories.
On the other hand, the Type II models are also useful. Though we know lit-
tle about their nature, there are such things as socialization effects. But nei-
ther Type I nor Type II models can claim to be valid generally. Becker (1996)
makes a good point when he writes that RCT can be scientifically more accept-
able than theories explaining behaviour by cultural, psychological or biologi-
cal forces. But he is wrong when he makes this statement general and also
when he assumes that we would be faced to the dilemma of choosing between
Type I and Type II models. In the provocative vocabulary of Amartya Sen, we
are not condemned to see the homo sociologicus either as a «rational idiot» being
rational exclusively in the instrumental sense or as an «irrational idiot» led by
forces beyond his control.
I will leave aside the eclectic writers who think they can solve the dilem-
ma at a low cost by assuming that people are rational in the sense of the type
I model when they choose the appropriate means to reach their objectives, but
irrational in the sense of the type II model when they select an objective or
follow a norm.
Papers 80 001-312 13/12/06 10:53 Página 160

160 Papers 80, 2006 Raymond Boudon

Type III model

By difference with the behavioural theory of Type I, the Type III theory con-
siders in Schütz’s vocabulary Weil-motive beside Wozu-motive. By difference
with Type II, it uses only marginally the notion of forces. In a nutshell, its
main difference with the other two models lies in the fact that it uses ordinary
psychology, rational theory in Nisbet’s sense, as its favourite theory of behav-
iour.
Before discussing the model from a theoretical viewpoint in more detail,
I will illustrate it. I will examine carefully an example from Durkheim. This
will have the incidental advantage of stressing, against a current view, that
Durkheim’s actual analyses do not use Type II, but rather Type III models. I
already mentioned this point when I recalled Durkheim’s analysis of the effect
of crisis situations on suicide rates. Hirschman (1982) has elaborated on the psy-
chological mechanism used by Durkheim in this analysis, using also a Type III
approach.
In a particularly brilliant analysis, Durkheim wonders why people execute
these particular actions which we call magical rituals. They do so because they
believe in the effectiveness of magical recipes. A difficult enigmatic question
on which many pages have been written is then: why do people believe that
rain rituals have the capacity of producing rain, while objectively they have
not. While many explanations use controversial psychological assumptions,
Durkheim succeeds solving the question by using rational ordinary psycho-
logy.
Magical rituals as rain rituals are instrumental, in the sense that they are
supposed to produce an effect, rain fall. Of course rain fall is wished because
of the effect of rain on the growth of plants. This instrumental side of magi-
cal practices raises no particular question. The puzzling aspect of this behav-
iour lies rather in the belief that magical rituals would be objectively useful.
An easy solution to this problem is to use a Type II model to explain the
belief: to assume for instance that the magician and his customers have been
socialized to believe that rain rituals are effective. This type of solution was
proposed notably by Lévy-Bruhl (1960 [1922]): he postulated the existence
of forces he christened «primitive mentality». They would make that the brain of
the «primitives», in the vocabulary of Lévy-Bruhl’s time, is submitted to infer-
ence rules different from ours. Lévy-Bruhl supposes in other words that ratio-
nality is culture dependent. The same type of solution was endorsed in mod-
ern days by Needham (1972), Beattie (1964) or Sahlins (1995). According to
the theories they developed, the rules of inference which we use and consider
universal would in fact be particular and characteristic of our own culture.
Huntington (1996) goes even as far in this vein as to say that the very notion
of universality is a particular feature of Western culture.
With his deep sense of methodology, Wittgenstein (1967) saw that this
explanation of magical beliefs is unsatisfactory and proposes an alternative
one: that Westerners would wrongly assume that the primitives would believe
Papers 80 001-312 13/12/06 10:53 Página 161

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 161

in the effectiveness of magic. According to Wittgenstein, magical behaviour

should be interpreted as expressive rather than instrumental: the «primitives»
would not really believe in the efficiency of their rituals. They would rather
express by their rituals their desire that rain falls.
Wittgenstein’s theory is definitely more palatable than the theory devel-
oped by Lévy-Bruhl and his followers. It uses ordinary psychology and belongs
to our Type III. Wittgenstein was conscious that the widespread Type II the-
ories of magical rituals were scientifically unsatisfactory. Unfortunately, his
theory is false in the sense that people reject with force the idea that magical rit-
uals would have no effect: they refuse the expressive interpretation of their
magical beliefs. According to Evans Pritchard (1968[1937]), when a Zande
falls on a root in the bush, he shows when interviewed that he knows that his
fall is the effect of mechanical forces, but he sees it at the same time as the
effect of «magical» factors. He knows, as we do, that he would plausibly not
have fallen if he had been more careful, but also if he would have been more
«lucky». As Evans Pritchard says, this combination of objective and magical
explanations is common to the Azande and to Westerners: we also explain
some states of affairs by luck. Evens Pritchard uses here ordinary psychology.
So, «primitive» people are convinced that their rituals have an instrumen-
tal value. To be able to endorse Wittgenstein’s answer, we should be ready to
assume that the «primitives» do not believe effectively what they say they
believe. But in that case we would have to explain this false consciousness.
This is more easily said than done. Moreover, Wittgenstein’s theory is incom-
patible with many observations, as Horton’s (1993, 1982): in the English
speaking countries of black Africa where he had conducted his investigations,
he observed that people believe in magical practices even when they have con-
verted to Christianity. When asked why, they answer that Christianity is good
from a soteriological viewpoint, but that it has the defect of missing the mag-
ical remedies offered by the magical rituals derived from the animistic religion
from which they came.
Both Weber and Durkheim and their followers, as Evans Pritchard or
Horton, propose a Type III solution to the problem of the explanation of mag-
ical rituals which is much more acceptable than its competitors.
Weber (1976 [1922]) encapsulates the theory he defends by a compact
sentence: «to the primitive fire making is as magical as rain making». This
means that to us Westerners, there is a difference between fire- and rain mak-
ing. The operation thanks to which the primitive makes fire, rubbing two
pieces of wood, is grounded on objectively valid laws. By contrast, we see rain
rituals as magical because they are not objectively grounded. To them, to the
«primitives», the distinction is meaningless: they see fire and rain making as
equally magical. Weber means here that we make a difference because we know
the laws of the transformation of energy. That the primitives have not learnt
them is clear. That they would have discovered them intuitively is implausi-
ble. In other words, the «primitives» have probably discovered empirically the
way of making fire, but the theory thanks to which they explain this technique
Papers 80 001-312 13/12/06 10:53 Página 162

162 Papers 80, 2006 Raymond Boudon

is likely as magical as the theory supporting their rain making techniques.

There is no reason to suppose with Lévy-Bruhl that the brain of the «primi-
tives» is wired in a way different from ours. It suffices to take seriously the
obvious fact that they have not learnt physics on school benches, and that they
have no reason to see intuitively laws that Westerners have discovered after
many centuries. There is no reason for them to make a difference between the
way they explain fire making and rain making, while for us there is a strong
reason.
Durkheim did not know Weber’s sentence, but he elaborates on the same
ideas. When the «primitives» grow some plant, they use much empirical know-
how which has been discovered and transmitted from one generation to the
next. But they also need to know why plants grow, wither and die. This can-
not be determined empirically. So they need to forge some «biological theo-
ry». As they have no access to science in the modern sense, they draw this bio-
logical theory from the religious interpretation of the world treated as legitimate
in their society. For religion represents the legitimate source of knowledge, as
science in Western societies. As to the magical recipes, they are technical pro-
cedures derived from this religious theory of the world. So, magical recipes are
comparable to the technical recipes Westerners draw from science. The obvi-
ous difference is that the recipes derived from science are more reliable. Does
not this show, as Lévy-Bruhl postulates, that the brain of the primitives is wired
otherwise; that they are not, say, as sensitive as Westerners to contradiction?
Not only, objects Durkheim, they dislike contradiction, but they treat it
as any modern scientist does: by inventing auxiliary assumptions. We know
from the Duhem-Quine thesis that, when a theory fails to explain some new
data, the normal reaction of any scientist is to invent auxiliary assumptions
rather than to reject the theory, because the theory explains many data and
also because he does not know a priori which element in the theory is wrong,
so that it is reasonable for him to assume that an auxiliary assumption will
likely reconcile the theory with the data. This is what the scientists do, as the
history of science shows. This is what they should do, as epistemology shows.
This is also what the magician does. In the case where the magical rituals fail,
he will assume, say, that they have not been executed exactly as they should
have been, etc. Durkheim anticipated very clearly the Duhem-Quine thesis,
a thesis widely accepted by historians and philosophers of science.
Another objection raised by Durkheim himself is that, being unground-
ed, magical recipes fail in 50% of the cases. Durkheim’s answer is that, as the
rain rituals are practised in the period of the year where rain is more likely to
fall, a correlation between the two variables days with/without rituals and days
with/without rain will be normally observed. The correlation will of course be
spurious. But Westerners ground also often their beliefs on spurious correlations.
On the whole, Durkheim’s theory is much more acceptable than the alter-
native theories of magical rituals available on the market. Moreover, it explains
easily and convincingly a number of puzzling data: why magical practices were
much more frequent in Europe in the 16th or 17th centuries than in the 14th cen-
Papers 80 001-312 13/12/06 10:53 Página 163

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 163

tury, and more frequent in the modern parts of Europe: more frequent in
Northern than in Southern Italy or than in Spain, say. It explains on the whole
many comparative data, some of them have been discovered a long time after
Durkheim proposed his theory, for instance by Thomas (1973). On the
whole, Durkheim’s Type III theory works beautifully where several Type II
theories have failed.
Durkheim uses here ordinary rational psychology. The primitive people
need recipes to grow their crops. Durkheim’s central thesis is that they are
rational in the same way as scientists are. To the best of their knowledge, they
have strong reasons to believe in the effectiveness of their magical recipes.
Evans Pritchard’s (1968 1937) or Horton’s (1993) work are fascinating
because they show also, in agreement with Durkheim’s intuition, that the primi-
tives use the same inference rules as any scientist. When they take their oracles,
the Azande manipulate logical calculus with a greater virtuosity than any western
student. Cole and Scribner (1974) have shown that in Liberia, the «primitives»
master the rules of logical inference as well as any American student.
I have tried to show elsewhere, taking several examples from classical and
modern sociology, that Type III models are used in many illuminating analy-
ses (Boudon, 1998a, 2003, 2005). The greatest sociologists, Tocqueville, as
well as Durkheim or Weber and others, the greatest anthropologists, Evans
Pritchard and others, always use Type III models. It is exceptional to see them
satisfied with the explanation that some phenomenon is due to the fact that
individuals are guided by psychological, social or biological forces

Type III model and Weber’s methodological writings

I borrowed my main example from Durkheim as a kind of provocation. In his
Elementary forms of religious life little remains from what he had stated in the
Rules of sociological method and generally in his methodological writings. In
the latter he prohibits any psychological explanation of sociological facts as the
famous motto states: explain social facts by social facts («expliquer les faits soci-
aux par des faits sociaux»). In his explanation of variations in the rates of sui-
cide, he always uses ordinary rational psychology though, as in his explana-
tion of the reasons as to why Protestant commit suicide more often than
Catholics, bachelors than married people, etc. The same is true of his analysis
of magical rituals. In all these analyses, he explains the social phenomena he
explores by making them the outcome of individual actions which he treats
as understandable in Weber’s sense: as explainable by the kind of reasons and
motivations which we evoke normally in everyday life. For various reasons
which I do not intend to explore here, his methodological writings have little
to do with the methodology he uses effectively in his analyses.
This discrepancy does not appear with Weber, though his methodological
writings are sketchy. When they are made less sketchy, we get the Type III
model illustrated by the examples I have just presented, which, by difference
with Type I and II models, can be considered as genuinely general.
Papers 80 001-312 13/12/06 10:53 Página 164

164 Papers 80, 2006 Raymond Boudon

So, let us make Weber’s methodological ideas more explicit and articula-
ted with one another.
The idea of comprehension means that we can in principle find the causes
of any piece of behaviour or of any belief, even if at first sight it appears as
strange to us, as «irrational» in the current sense of the word. This is the case
for instance of magical beliefs. They give an impression of being strongly irra-
tional. Still, as Durkheim and Weber have shown, they can be convincingly
explained in the sense that their causes can be identified in a convincing way.
These causes are the reasons and motivations people have to believe what they
believe. The motivations in this case are clear: to have a sufficient crop to be
able to survive. The reasons why they endorse the idea that magical rituals have
been identified by Durkheim and Weber. The «primitives» have at their disposal
theories of the world from which they draw magical recipes, exactly as we
extract technical recipes from our scientific theories. They confront their the-
ories to the data and eliminate eventual contradictions using the same devices.
So, to explain an action, even apparently irrational as rain rituals, means recon-
structing the motivations and reasons of the actors.
Another point of the comprehension theory is that reconstructing the
reasons and motivations of an actor amounts to building a theory following
the rules generally used when building any theory. Weber’s example of the
wood cutter makes this point clear. If I see somebody cutting wood in his
yard, I will explain his behaviour by evoking some possible motivations and
reasons. Maybe he is cold and wants to burn the wood in his chimney. This
theory will be rejected if the weather is warm and I will have to find some
other one, until I reach a theory lending to the actor motivations and rea-
sons borrowed from the ordinary rational psychology and compatible with
all the available observational data. In the case where I would fail to find
understandable motivations and reasons, and only in this case, I would legi-
timately propose an irrational explanation as: he suffers from a compulsive
trend to cut wood.
Another very important point which can be derived from Weber’s notes is
that the expression «being rational» can be treated as a shorthand notation
meaning actually: «behaving, acting or believing on the basis of a set of rea-
sons perceived as strong and well articulated with one another». The example
from Durkheim on magical rituals provides a clear illustration of this point:
the magical rituals are explained by a set of articulated reasons in the mind
of the «primitives».
Another important point is that some of the reasons explaining an action
can be instrumental and some not. The very distinction made by Weber
between instrumental and axiological rationality shows that, to him, rationali-
ty does not necessarily take an instrumental form.
The meaning of this notion of axiological rationality has been much dis-
cussed. Sukale (1995) went even as far as to say that it is meaningless. My own
interpretation is that we can and have an interest to distinguish between instru-
mental rationality and cognitive rationality. Instrumental rationality means:
Papers 80 001-312 13/12/06 10:53 Página 165

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 165

looking for the best means to reach a goal. Cognitive rationality means: finding
the good explanation of a phenomenon. As to axiological rationality, I pro-
pose to see it as a special case of cognitive rationality. If this interpretation
—which is not to be found in Weber’s work but which can be regarded as
elaborating on his intuitions- is accepted, being axiologically rational means:
finding a set of strong and strongly articulated reasons leading to a normative
conclusion. I have tried elsewhere to illustrate this idea by an example taken from
Adam Smith, where he wonders why the British public in his time was con-
vinced that miners should be definitely paid more than soldiers. He explains
this puzzling collective belief by showing that this conclusion was derived in the
mind of the public by a set of strong reasons, strongly related to one another:
a brilliant example of «axiological rationality» (Boudon, 1998b). I will evoke
contemporary illustrations in a moment.
A further important point is that the resources of the so-called deep psy-
chology should not be mobilized because the forces thanks to which it pro-
poses to explain behaviour are in most cases conjectures which cannot be test-
ed, so that using deep psychology takes sociological theories away from the
realm of scientific theories.
The above principles lead many of the most convincing and illuminating
analyses we owe to sociologists. Most of them use in other words the behavioural
theory of Type III. All recognize that instrumental rationality appears in com-
bination with non instrumental rationality, the rationality I called cognitive, a
special case being what Weber christened axiological rationality.
Also, it can be shown that the phenomena which appear as unexplainable
in the frame of the behavioural theory of Type I are in most cases explain-
able by the behavioural theory of Type III. Frey’s paradox can be solved if we
see that, if no compensation is proposed, people will accept more easily the
nuclear waste on the territory of their community, because they tend to devel-
op the theory that they will consider themselves and will be considered as able
to accept a sacrifice for the sake of the general interest. By contrast, when com-
pensation is proposed, people tend to develop the theory that a bargain is pro-
posed to them and that they feel entitled to judge that the bargain is actually
not very interesting. Because the proposal generates two theories in the mind
of people, those to whom compensation is offered accept the proposal less fre-
quently. That the same observation was made in two different contexts tends
to show that the proposal tended to induce the same system of reasons in the
people’s minds in the two contexts.
I have tried to show that the findings of cognitive psychology generally
explained by Type II models could also easily be explained by Type III mod-
els: one need not to postulate that the wrong answers are produced by biases
with a highly conjectural origin (Boudon, 1996). On the whole, it can be
shown both by the a priori arguments I have tried to develop, but also a pos-
teriori by actual analyses of empirical data, that the difficulties met by the
behavioural theory of Type I and II can in many cases be solved by using
the behavioural theory of Type III.
Papers 80 001-312 13/12/06 10:53 Página 166

166 Papers 80, 2006 Raymond Boudon

The importance of the discussion

To conclude, I will point to the fact that the former discussion is important
not only from a scientific but also from a general viewpoint.
Suppose we want to explain why most of us have a negative reaction toward
female genital mutilation. An explanation using the behavioural theory of Type
II would assume that cultural forces and socialization effects make that we
have a feeling of cruelty, while other cultural forces make that this practice was
and still is widely adopted in many societies without arousing the same feeling.
Such explanations are proposed by important writers as Shweder (2000) or
Geertz (1984). It is difficult to imagine an explanation of this difference which
would use the behavioural theory of Type I. An explanation using the behav-
ioural theory of Type III would assume, say, that female genital mutilation is
a means of social control and that as soon as a means of social control less cruel
than another is found we tend to require that the less cruel means is adopted.
Durkheim (1960 [1893]) developed this kind of explanation when he made the
point that social control tends to become over time more and more liberal.
Tocqueville (1986 [1840]) developed the same kind of model when he remarked
that our moral sensibility toward slavery has irreversibly changed. «The great-
est minds in Greece and Rome, he states, never imagined that slavery could
be abolished». They accepted it because they thought it was functional: indis-
pensable to the economic system. Then, forms of social organization appeared
which did not have recourse to slavery. From that moment a negative feeling
toward slavery developed. In the 18th century, Montesquieu wrote that it was
impossible to abolish slavery in the Antilles because this would deeply disturb
the sugar market, but he considered it as «against nature». In other words, the
feelings and reactions toward an institution result from a set of articulated rea-
sons. The Greek saw slavery as indispensable. It was no more possible to endorse
this view once it had been shown that it was not. Still at the end of the 19th
century universal vote was considered as a dangerous institution, until it was
shown it was not. All these familiar examples illustrate the tension between
axiological and instrumental rationality. The political men at the end of the
18th century had strong axiological reasons to condemn slavery and strong
instrumental reasons not to abolish it.
I will evoke a final example to illustrate the importance of the cognitive
approach to rationality which characterizes the behavioural theory of Type III.
The example deals with differences in the moral reactions and sensitiveness of
the public to current events.
In an article of December 24, 2004, the International Herald Tribune rais-
es an interesting sociological question. It wonders why very serious violations
of the law can arouse little reaction in a given context while much less serious
violations can generate more reprobation in another. The article reports that «in
Germany, the deputy police chief of Frankfurt was found guilty of threatening
to torture the captured kidnapper-murderer of an 11-year-old child», though
he «had every reason to believe that a kidnapped child’s life hung in the balance,
Papers 80 001-312 13/12/06 10:53 Página 167

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 167

and, in this sense, threatening torture of the young man who had seized him
was justified». By contrast, while the US has used torture at the Abu Ghraib
prison in Baghdad and in Guantánamo, «there is no strong sign that the pub-
lic is outraged by these actions. The public reaction can be explained in the
two cases by reasons. But, as the German and the American contexts are dif-
ferent, the reasons were different. The decision of the German judge was explic-
itly justified by reference to the German past.» By contrast, «the American
public, victimized by a terrible act of mass murder and witness to the daily
terror bombings in Iraq, seems inured to the torture being conducted, sup-
posedly, in its interest.»
The same type of cognitive contextual rationality applies to the difference
in the treatment of Stalinism and Hitlerism. One can be an ex-Stalinist not
an ex-Nazi, because Stalinism is still analyzed by many minds as a perversion
of morally and socially acceptable ideas, by difference with Hitlerism which
is seen as inspired by devilish ideas.
Men are neither rational idiots nor irrational idiots. They behave in the
way they do because they have all kinds of theories on all kinds of questions.
In many cases they have to muddle through the complexity of the questions they
are faced to. Durkheim’s idea that the magician, the ordinary man and the sci-
entist behave in the same way is a deep idea. It sketches an important gener-
al theory of the homo sociologicus.
My final guess is that the social sciences are less exciting now than in
Tocqueville’s or Durkheim’s and Weber’s time because macroscopic sociology
has become impressionistic, while it had a definite scientific character in
Tocqueville’s, Durkheim’s or Weber’s work. And the main reason why they
produced convincing macroscopic theories is that they used a theory of behav-
iour of Type III rather than of Type I or II, the dominant theories of behaviour
today.

References
AXELROD, R. (1984). The Evolution of Cooperation. New York: Basic Books.
BEATTIE, J. (1964). Other Cultures. London: Cohen & West.
BECKER, G. (1996). Accounting for Tastes. Cambridge: Harvard U. Press.
BOUDON, R (1996). «The cognitivist model : a generalized Rational Choice model».
Rationality and Society, 8, 2, 123-150.
— (1998a). «Limitations of Rational Choice Theory». American Journal of Sociology,
104, 3, 817-828. Reproduced in: TERRELL & MEIER (eds.). Readings in Deviant
Behavior. Harcourt College Publishers, Fort Worth, Tx., USA.
— (1998b). «Social mechanisms without black boxes». In: HEDSTRÖM, Peter;
SWEDBERG, Richard (eds.). Social mechanisms: an analytical approach to social the-
ory. Cambridge: Cambridge University Press, 172-203.
— (2003). «Beyond Rational Choice Theory». Annual Review of Sociology, 29, 1-21.
— (2005). Tocqueville for Today. Oxford: Bardwell.
BUCHANAN, J.; TULLOCK, G. (1962). The Calculus of Consent. Ann Arbor: University
of Michigan Press.
Papers 80 001-312 13/12/06 10:53 Página 168

168 Papers 80, 2006 Raymond Boudon

COLE, L.; SCRIBNER, S. (1974). Culture and Thought: a Psychological Introduction. New
York: Wiley.
COLEMAN, J. (1986). Individual Interests and Collective Action: Selected Essays. Cambridge:
Cambridge U. Press.
DAMASIO, A.R. (1994). Descartes’ Error: Emotion, Reason and the Human Brain. New
York: Grosset/Putnam.
— (2003). Looking for Spinoza: Joy, Sorrow and the Feeling Brain. Harcourt.
DAWKINS, R. (1976). The Selfish Gene. Oxford: Oxford University Press.
DURKHEIM, E. (1960 [1893]). De la division du travail social. Paris: Presses Universitaires
de France.
— (1967[1897]). Le suicide, étude sociologique. Paris: Presses Universitaires de
France.
— (1979 [1912]). Les Formes élémentaires de la vie religieuse. Paris: Presses Universitaires
de France.
EVANS-PRITCHARD, E.E. (1968[1937]). Witchcraft, Oracles and Magic among the Azande.
Oxford: Clarendon Press.
FEREJOHN, J.A.;FIORINA, M.P. (1974). «The paradox of not voting: A decision theo-
retic analysis». The American political science review, 68, 2, juin, 525-36.
FREY, B.S. (1997). Not just for the money: an economic theory of personal motivation.
Cheltenham: Edward Elgar.
GEERTZ, C. (1984). «Distinguished Lecture: Anti anti-relativism». American anthro-
pologist, vol. 86, n°2, 263-278.
HIRSCHMAN, A.O. (1982). Shifting Involvements. Private interests and Public Action.
Princeton: Princeton University Press.
HORTON, R. (1993). «Lévy-Bruhl, Durkheim and the Scientific Revolution». In:
HORTON, R. Patterns of Thought in Africa and the West. Cambridge: Cambridge
University Press, 63-104.
— (1982). «Tradition and Modernity Revisited». In HOLLIS, M.; LUKES, S. Rationality
and Relativism. Oxford: Blackwell, 201-60.
HUNTINGTON, S. (1996). The Clash of Civilizations and the Remaking of the World
Order. New York: Shuster and Shuster.
LÉVY-BRUHL, L. (1960 [1922]). La Mentalité primitive. Paris: Presses Universitaires
de France.
MOULIN, L. (1953). «Les origines religieuses des techniques électorales et délibératives
modernes». Revue internationale d’histoire politique et constitutionnelle, avril-juin,
106-148.
NEEDHAM, R. (1972). Belief, Language and Experience. Oxford: Blackwell.
NISBET, R. (1966). The Sociological Tradition. Glencoe, Ill.: The Free Press.
OVERBYE, E. (1995). «Making a case for the rational, self-regarding,“‘ethical” voter…
and solving the “Paradox of not voting” in the process». European Journal of Political
Research, 27, 369-96.
POPKIN, S.(1979). The Rational Peasant. The Political Economy of Rural Society in
Vietnam. Berkeley: University of California Press.
POPPER, K. (1976). «The Myth of the Framework». In FREEMAN, E. (ed.). The
Abdication of Philosophy : Philosophy and the Public Good. La Salle, Ill.: Open Court,
p. 23-48.
RADNITZKY, G. (1987). «La perspective économique sur le progrès scientifique: appli-
cation en philosophie de la science de l’analyse coût-bénéfice». Archives de philo-
sophie, 50, avril-juin, 177-198.
Papers 80 001-312 13/12/06 10:53 Página 169

Homo Sociologicus: Neither a Rational nor an Irrational Idiot Papers 80, 2006 169

SAHLINS, M. (1995). How «Natives» Think. About Captain Cook for instance. Chicago:
Chicago U. Press.
SHWEDER, R. A. (1977). «Likeliness and Likelihood in Everyday Thought: Magical
Thinking in Judgments about Personality». Current Anthropology, vol.18, n°4,
décembre, 637-659.
— (2000). «What about “female genital mutilation” and why understanding culture
matters in the first place». Daedalus, 129, 4, 209-232.
SIMMEL, G. (1892). Die Probleme der Geschichtsphilosophie, München: Duncker &
Humblot.
SUKALE, S. (1995). «Introduction to Max Weber». Schriften zur Soziologie. Stuttgart:
Reclam.
THOMAS, K. (1973). Religion and the Decline of Magic. Harmondsworth: Penguin
Books.
TOCQUEVILLE, A. de (1986 [1840]). La Démocratie en Amérique, vol. II. In Tocqueville,
De la démocratie en Amérique, Souvenirs, L’Ancien Régime et la Révolution. Paris:
Robert Laffont, Bouquins.
TOOBY, J.; COSMIDES, L. (1992). «The psychological foundations of culture». In:
BARKOW, J.; COSMIDES, L.; TOOBY, J. (eds.). The adapted mind: Evolutionary psy-
chology and the generation of culture. New York: Oxford University Press.
WEBER, M. (1976 [1922]). Wirtschaft und Gesellschaft. Tübingen, Mohr, 3 vol.
— (1988 [1920-21]). Gesammelte Aufsätze zur Religionssoziologie. Tübingen, Mohr,
3 vol.
WITTGENSTEIN, L. (1967). «Bemerkungen über Frazer’s The Golden Bough». Synthese,
17, 233-253.